Strain-Level Discrimination of Shiga Toxin-Producing Escherichia coli in Spinach Using Metagenomic Sequencing

Abstract

Consumption of fresh bagged spinach contaminated with Shiga toxin-producing Escherichia coli (STEC) has led to severe illness and death; however current culture-based methods to detect foodborne STEC are time consuming. Since not all STEC strains are considered pathogenic to humans, it is crucial to incorporate virulence characterization of STEC in the detection method. In this study, we assess the comprehensiveness of utilizing a shotgun metagenomics approach for detection and strain-level identification by spiking spinach with a variety of genomically disparate STEC strains at a low contamination level of 0.1 CFU/g. Molecular serotyping, virulence gene characterization, microbial community analysis, and E. coli core gene single nucleotide polymorphism (SNP) analysis were performed on metagenomic sequence data from enriched samples. It was determined from bacterial community analysis that E. coli, which was classified at the phylogroup level, was a major component of the population in most samples. However, in over half the samples, molecular serotyping revealed the presence of indigenous E. coli which also contributed to the percent abundance of E. coli. Despite the presence of additional E. coli strains, the serotype and virulence genes of the spiked STEC, including correct Shiga toxin subtype, were detected in 94% of the samples with a total number of reads per sample averaging 2.4 million. Variation in STEC abundance and/or detection was observed in replicate spiked samples, indicating an effect from the indigenous microbiota during enrichment. SNP analysis of the metagenomic data correctly placed the spiked STEC in a phylogeny of related strains in cases where the indigenous E. coli did not predominate in the enriched sample. Also, for these samples, our analysis demonstrates that strain-level phylogenetic resolution is possible using shotgun metagenomic data for determining the genomic relatedness of a contaminating STEC strain to other closely related E. coli.

Fig 1. Microbial community associated with enriched STEC-spiked spinach.

Spinach samples were spiked with STEC at a level of 0.1 CFU/g and enriched. Triplicate experiments for each strain were performed using bagged spinach purchased at different times and designated m1, m2, and m3. The E. coli phylogroup of the spiked STEC strain is indicated at the top of the figure. Bacterial genera or species contributing ≥1% to the total identified population in at least one sample are shown and the sum of all other species identified in lower abundances is included in “other”. Samples in which the serogroup of the spiked STEC was the only serogroup detected (green asterisk) and samples for which the serotype or virulence genes were not detected (red asterisk) are indicated.

Fig 2. Phylogenetic relationships resulting from E. coli core gene SNP analysis of spiked STEC genomes, metagenomic samples, and other E. coli strains.

This neighbor-joining tree was constructed using a p distance matrix and 500 bootstrap replications. Metagenomic samples are color-coded as follows: green, spiked STEC only E. coli detected in sample; blue, indigenous E. coli present but clusters with spiked strain; and red, indigenous E. coli present and does not cluster with spiked strain. Closed genome strains are labeled with their strain name and GenBank accession number. Phylogroups are indicated by the square brackets. Bootstrap values >80% are given at the internal nodes.

Fig 3. Finer-scale resolution of the phylogenetic relationships resulting from E. coli core gene SNP analysis.

Neighbor-joining trees for the O157:H7 and O77:H18 clonal groups were constructed using a p distance matrix and 500 bootstrap replications. The spiked STEC strains and metagenomic samples are indicated by the black and gray circles, respectively. Bootstrap values >80% are given at the internal nodes.

Fig 4. Finer-scale resolution of the phylogenetic relationships resulting from E. coli core gene SNP analysis.

Neighbor-joining trees for the O91:H21, O104:H4, O121:H19, O26:H11, O145:H28, and O113:H21 clonal groups were constructed using a p distance matrix and 500 bootstrap replications. The spiked STEC strains and metagenomic samples are indicated by the black and gray circles, respectively. Bootstrap values >80% are given at the internal nodes.